Automatic multi-resolution shape modeling of multi-organ structures.

Point Distribution Models (PDM) are among the most popular shape description techniques and their usefulness has been demonstrated in a wide variety of medical imaging applications. However, to adequately characterize the underlying modeled population it is essential to have a representative number of training samples, which is not always possible. This problem is especially relevant as the complexity of the modeled structure increases, being the modeling of ensembles of multiple 3D organs one of the most challenging cases. In this paper, we introduce a new GEneralized Multi-resolution PDM (GEM-PDM) in the context of multi-organ analysis able to efficiently characterize the different inter-object relations, as well as the particular locality of each object separately. Importantly, unlike previous approaches, the configuration of the algorithm is automated thanks to a new agglomerative landmark clustering method proposed here, which equally allows us to identify smaller anatomically significant regions within organs. The significant advantage of the GEM-PDM method over two previous approaches (PDM and hierarchical PDM) in terms of shape modeling accuracy and robustness to noise, has been successfully verified for two different databases of sets of multiple organs: six subcortical brain structures, and seven abdominal organs. Finally, we propose the integration of the new shape modeling framework into an active shape-model-based segmentation algorithm. The resulting algorithm, named GEMA, provides a better overall performance than the two classical approaches tested, ASM, and hierarchical ASM, when applied to the segmentation of 3D brain MRI.

Fully automatic GBM segmentation in the TCGA-GBM dataset: Prognosis and correlation with VASARI features.

Reproducible definition and quantification of imaging biomarkers is essential. We evaluated a fully automatic MR-based segmentation method by comparing it to manually defined sub-volumes by experienced radiologists in the TCGA-GBM dataset, in terms of sub-volume prognosis and association with VASARI features. MRI sets of 109 GBM patients were downloaded from the Cancer Imaging archive. GBM sub-compartments were defined manually and automatically using the Brain Tumor Image Analysis (BraTumIA). Spearman's correlation was used to evaluate the agreement with VASARI features. Prognostic significance was assessed using the C-index. Auto-segmented sub-volumes showed moderate to high agreement with manually delineated volumes (range (r): 0.4 - 0.86). Also, the auto and manual volumes showed similar correlation with VASARI features (auto r = 0.35, 0.43 and 0.36; manual r = 0.17, 0.67, 0.41, for contrast-enhancing, necrosis and edema, respectively). The auto-segmented contrast-enhancing volume and post-contrast abnormal volume showed the highest AUC (0.66, CI: 0.55-0.77 and 0.65, CI: 0.54-0.76), comparable to manually defined volumes (0.64, CI: 0.53-0.75 and 0.63, CI: 0.52-0.74, respectively). BraTumIA and manual tumor sub-compartments showed comparable performance in terms of prognosis and correlation with VASARI features. This method can enable more reproducible definition and quantification of imaging based biomarkers and has potential in high-throughput medical imaging research.

Quantitative multi-slice computed tomography assessment of the mitral valvular complex for transcatheter mitral valve interventions part 1: systematic measurement methodology and inter-observer variability

AIMS Transcatheter mitral valve replacement (TMVR) is an emerging technology with the potential to treat patients with severe mitral regurgitation at excessive risk for surgical mitral valve surgery. Multimodal imaging of the mitral valvular complex and surrounding structures will be an important component for patient selection for TMVR. Our aim was to describe and evaluate a systematic multi-slice computed tomography (MSCT) image analysis methodology that provides measurements relevant for transcatheter mitral valve replacement. METHODS AND RESULTS A systematic step-by-step measurement methodology is described for structures of the mitral valvular complex including: the mitral valve annulus, left ventricle, left atrium, papillary muscles and left ventricular outflow tract. To evaluate reproducibility, two observers applied this methodology to a retrospective series of 49 cardiac MSCT scans in patients with heart failure and significant mitral regurgitation. For each of 25 geometrical metrics, we evaluated inter-observer difference and intra-class correlation. The inter-observer difference was below 10% and the intra-class correlation was above 0.81 for measurements of critical importance in the sizing of TMVR devices: the mitral valve annulus diameters, area, perimeter, the inter-trigone distance, and the aorto-mitral angle. CONCLUSIONS MSCT can provide measurements that are important for patient selection and sizing of TMVR devices. These measurements have excellent inter-observer reproducibility in patients with functional mitral regurgitation.

MASCG: Multi-Atlas Segmentation Constrained Graph method for accurate segmentation of hip CT images

This paper addresses the issue of fully automatic segmentation of a hip CT image with the goal to preserve the joint structure for clinical applications in hip disease diagnosis and treatment. For this purpose, we propose a Multi-Atlas Segmentation Constrained Graph (MASCG) method. The MASCG method uses multi-atlas based mesh fusion results to initialize a bone sheetness based multi-label graph cut for an accurate hip CT segmentation which has the inherent advantage of automatic separation of the pelvic region from the bilateral proximal femoral regions. We then introduce a graph cut constrained graph search algorithm to further improve the segmentation accuracy around the bilateral hip joint regions. Taking manual segmentation as the ground truth, we evaluated the present approach on 30 hip CT images (60 hips) with a 15-fold cross validation. When the present approach was compared to manual segmentation, an average surface distance error of 0.30 mm, 0.29 mm, and 0.30 mm was found for the pelvis, the left proximal femur, and the right proximal femur, respectively. A further look at the bilateral hip joint regions demonstrated an average surface distance error of 0.16 mm, 0.21 mm and 0.20 mm for the acetabulum, the left femoral head, and the right femoral head, respectively.

Ventilation/perfusion SPECT or SPECT/CT for lung function imaging in patients with pulmonary emphysema?

PURPOSE To evaluate the utility of attenuation correction (AC) of V/P SPECT images for patients with pulmonary emphysema. MATERIALS AND METHODS Twenty-one patients (mean age 67.6 years) with pulmonary emphysema who underwent V/P SPECT/CT were included. AC/non-AC V/P SPECT images were compared visually and semiquantitatively. Visual comparison of AC/non-AC images was based on a 5-point likert scale. Semiquantitative comparison assessed absolute counts per lung (aCpLu) and lung lobe (aCpLo) for AC/non-AC images using software-based analysis; percentage counts (PC = (aCpLo/aCpLu) × 100) were calculated. Correlation between AC/non-AC V/P SPECT images was analyzed using Spearman's rho correlation coefficient; differences were tested for significance with the Wilcoxon rank sum test. RESULTS Visual analysis revealed high conformity for AC and non-AC V/P SPECT images. Semiquantitative analysis of PC in AC/non-AC images had an excellent correlation and showed no significant differences in perfusion (ρ = 0.986) or ventilation (ρ = 0.979, p = 0.809) SPECT/CT images. CONCLUSION AC of V/P SPECT images for lung lobe-based function imaging in patients with pulmonary emphysema do not improve visual or semiquantitative image analysis.

Pollen and charcoal in lake sediments compared with historically documented forest fires in southern Switzerland since AD 1920

Charcoal in unlaminated sediments dated by 210Pb was analysed by the pollen-slide and thin-section methods. The results were compared with the number and area of forest fires on different spatial scales in the area around Lago di Origlio as listed in the wildfire database of southern Switzerland since AD 1920. The influx of the number of charcoal particles > 75 µm2 in pollen slides correlates well with the number of annual forest fires recorded within a distance of 20-50 km from the coring site. Hence a size-class distinction or an area measurement by image analysis may not be absolutely necessary for the reconstruction of regional fire history. A regression equation was computed and tested against an independent data set. Its use makes it possible to estimate the charcoal area influx (or concentration) from the particle number influx (or concentration). Local fires within a radius of 2 km around the coring site correlate well with the area influx of charcoal particles estimated by the thin-section method measuring the area of charcoal particles larger than 20 000 µm2 or longer than 50 µm. Pollen percentages and influx values suggest that intensive agriculture and Castanea sativa cultivation were reduced 30-40 years ago, followed by an increase of forest area and a development to more natural woodlands. The traditional Castanea sativa cultivation was characterized by a complete use of the biomass produced, so abandonment of chestnut led to an increasing accumulation of dead biomass, thereby raising the fire risk. On the other hand, the pollen record of the regional vegetation does not show any clear response to the increase of fire frequency during the last three decades in this area.

Dish Detection and Segmentation for Dietary Assessment on Smartphones

Diet-related chronic diseases severely affect personal and global health. However, managing or treating these diseases currently requires long training and high personal involvement to succeed. Computer vision systems could assist with the assessment of diet by detecting and recognizing different foods and their portions in images. We propose novel methods for detecting a dish in an image and segmenting its contents with and without user interaction. All methods were evaluated on a database of over 1600 manually annotated images. The dish detection scored an average of 99% accuracy with a .2s/image run time, while the automatic and semi-automatic dish segmentation methods reached average accuracies of 88% and 91% respectively, with an average run time of .5s/image, outperforming competing solutions.

High intratumoral levels of FoxP3+ lymphocytes are associated with better prognosis in primary resected esophageal adenocarcinomas.

Background: Tumor infiltrating T-lymphocytes (TILs) have been shown to play an important prognostic role in many carcinomas. The identification of prognostic relevant morphological or molecular factors is a major area of interest in the diagnostic process and for the treatment of highly aggressive esophageal adenocarcinoma. Studies about the impact of TILs in this tumor have not shown completely congruent results yet. We present a comprehensive study about the clinical and pathological impact of TIL in esophageal adenocarcinomas. Methods: A next generation tissue microarray (TMA) of 117 primary resected esophageal adenocarcinomas was analyzed for CD3+, CD8+ and FoxP3+ TIL using immunohistochemistry. The TMA contained three cores of the tumor center and the tumor periphery per each case. Slides were scanned with a high-resolution scanner (ScanScope CS; Aperio) and an image analysis software (Aperio Image Scope) was used to determine the TIL counts. The results were correlated with clinicopathological parameters. Results: CD3+, CD8+ and FoxP3+ TIL counts showed a significant correlation among each other (p<0.001 each, range: 0.27-0.77). TIL counts were categorized as high and low levels, according to the median. Tumors with high FoxP3+ intratumoral lymphocyte counts were more frequently of lower pT category (p<0.001) and without lymph node metastasis (p=0.04). High levels of FoxP3+ lymphocytes in the tumor center and the periphery were also associated with better prognosis (p<0.001 and p=0.041, respectively) in univariate analysis. A similar prognostic impact was seen for high levels of CD3+ and CD8+ TIL in the tumor center, but not in the periphery (p=0.047 and p=0.011, respectively). In multivariate analysis high central FoxP3+TIL levels were an independent prognostic factor (HR=0.4; p=0.023) which was similar to a combination score of CD3+/CD8+/FoxP3+ TIL (HR=0.54; p=0.027) or CD8+/Foxp3+ TIL (HR=0.052; p=0.020) and superior to pT- and pN category (p>0.05 each). Conclusion: This study demonstrates a significant beneficial prognostic impact of high TIL counts in the tumor center of esophageal adenocarcinomas, in particular with regards to the subpopulation of FoxP3+ and CD8+ T-regulatory cells. The determination of intratumoral lymphocytic counts and application of TIL scores can improve prognostic accuracy of pathologic reports of these tumors and may be helpful for better risk stratification of esophageal adenocarcinoma patients.

Lung Pattern Classification for Interstitial Lung Diseases Using a Deep Convolutional Neural Network

Automated tissue characterization is one of the most crucial components of a computer aided diagnosis (CAD) system for interstitial lung diseases (ILDs). Although much research has been conducted in this field, the problem remains challenging. Deep learning techniques have recently achieved impressive results in a variety of computer vision problems, raising expectations that they might be applied in other domains, such as medical image analysis. In this paper, we propose and evaluate a convolutional neural network (CNN), designed for the classification of ILD patterns. The proposed network consists of 5 convolutional layers with 2×2 kernels and LeakyReLU activations, followed by average pooling with size equal to the size of the final feature maps and three dense layers. The last dense layer has 7 outputs, equivalent to the classes considered: healthy, ground glass opacity (GGO), micronodules, consolidation, reticulation, honeycombing and a combination of GGO/reticulation. To train and evaluate the CNN, we used a dataset of 14696 image patches, derived by 120 CT scans from different scanners and hospitals. To the best of our knowledge, this is the first deep CNN designed for the specific problem. A comparative analysis proved the effectiveness of the proposed CNN against previous methods in a challenging dataset. The classification performance (~85.5%) demonstrated the potential of CNNs in analyzing lung patterns. Future work includes, extending the CNN to three-dimensional data provided by CT volume scans and integrating the proposed method into a CAD system that aims to provide differential diagnosis for ILDs as a supportive tool for radiologists.

Carbohydrate Estimation by a Mobile Phone-Based System Versus Self-Estimations of Individuals With Type 1 Diabetes Mellitus: A Comparative Study

Background: Diabetes mellitus is spreading throughout the world and diabetic individuals have been shown to often assess their food intake inaccurately; therefore, it is a matter of urgency to develop automated diet assessment tools. The recent availability of mobile phones with enhanced capabilities, together with the advances in computer vision, have permitted the development of image analysis apps for the automated assessment of meals. GoCARB is a mobile phone-based system designed to support individuals with type 1 diabetes during daily carbohydrate estimation. In a typical scenario, the user places a reference card next to the dish and acquires two images using a mobile phone. A series of computer vision modules detect the plate and automatically segment and recognize the different food items, while their 3D shape is reconstructed. Finally, the carbohydrate content is calculated by combining the volume of each food item with the nutritional information provided by the USDA Nutrient Database for Standard Reference. Objective: The main objective of this study is to assess the accuracy of the GoCARB prototype when used by individuals with type 1 diabetes and to compare it to their own performance in carbohydrate counting. In addition, the user experience and usability of the system is evaluated by questionnaires. Methods: The study was conducted at the Bern University Hospital, “Inselspital” (Bern, Switzerland) and involved 19 adult volunteers with type 1 diabetes, each participating once. Each study day, a total of six meals of broad diversity were taken from the hospital’s restaurant and presented to the participants. The food items were weighed on a standard balance and the true amount of carbohydrate was calculated from the USDA nutrient database. Participants were asked to count the carbohydrate content of each meal independently and then by using GoCARB. At the end of each session, a questionnaire was completed to assess the user’s experience with GoCARB. Results: The mean absolute error was 27.89 (SD 38.20) grams of carbohydrate for the estimation of participants, whereas the corresponding value for the GoCARB system was 12.28 (SD 9.56) grams of carbohydrate, which was a significantly better performance ( P=.001). In 75.4% (86/114) of the meals, the GoCARB automatic segmentation was successful and 85.1% (291/342) of individual food items were successfully recognized. Most participants found GoCARB easy to use. Conclusions: This study indicates that the system is able to estimate, on average, the carbohydrate content of meals with higher accuracy than individuals with type 1 diabetes can. The participants thought the app was useful and easy to use. GoCARB seems to be a well-accepted supportive mHealth tool for the assessment of served-on-a-plate meals.

Methods for preparing and counting biochemical varves

Four different preparation and counting methods for biochemical varves were compared in order to assess counting errors and to standardize these techniques. The properties of two embedding methods, namely the shock-freeze, freeze-dry and the water-acetone-epoxy-exchange method, are discussed. Varve counts were carried out on fresh sediment and on sediment thin-sections, on the latter by manual and by automated counting using image-analysis software. Counting on fresh sediment and using image-analysis generally underestimated the number of varves, especially in sections with inconspicuous varves. A comparison between multiple varve counts carried out by a single analyst and different analysts showed no significant differences in the mean varve counts.